label looks like an autogenerated "whole disk" label, make sure that
we adjust the 'a' partition when we create an MBR that does not start
right at the start of the disk. Also when we force this partition to be
mounted, init it's mount point to / and force a newfs.
improved modularity and extensibility.
In the new architecture, a directed graph of argument-matching
objects (match objects) expresses the set of feasible ifconfig
statements. Match objects are labelled by subroutines that provide
the statement semantics.
Many IPv4, IPv6, 802.11, tunnel, and media configurations have been
tested.
AppleTalk, ISO, carp(4), agr(4), and vlan(4) configuration need
testing.
- Make MULTIPROCESSOR mandatory on i386.
Installation changes:
- Update installation section of release notes to match reality.
- Rename INSTALL to INSTALL_FLOPPY, retire INSTALL_LARGE.
- Build INSTALL kernel from GENERIC, like on amd64.
- Update boot menu to allow disabling ACPI and/or SMP.
- Remove GENERIC.NOACPI from the installed kernel list.
- TODO: install default boot.cfg in etc.tgz.
- TODO: possibly enable PCI fixup stuff at runtime if ACPI is disabled.
Build changes:
- No longer build ALL, it's for verification, is slow to build and the
build process is already crippled by the number of kernels built.
- No longer build GENERIC.NOACPI.
chose manual, the generated rc.conf has dhclient details instead
of defaultroute. (Noticed this when dhclient ran successfully,
but network failed so I entered network settings manually.)
Now only defines dhclient rc.conf variables when DHCP is used
else defines the defaultroute.
install time. Translations will need to be adjusted - I didn't feel
comfortable touching anything but .en and .de despite the relatively
simple adjustment...
at the end of `init_set_status'.
Previously, i386/amd64, mac68k and sgimips were using `md_init' to set the
appropriate kernel, but any changes made were being undone by a later call
to `init_set_status'.
This should fix IP3x installation issues on sgimips.
XXX - `md_init' is now unused. Should we G/C it?
all static. This makes it possible to write a md_make_bsd_partitions()
function without having to copy all of bsddisklabel.c into md.c.
Make ofppc sysinst install sanely and in a bootable manner on both my
7044, and my pegasos. The only thing the user needs to do is pick the
right kernel set.
with a suitable ramdisk etc, you can actually install successfully.
mostly the changes from port-prep are to switch the prep mkbootimage
to an msdos boot partition with a copy of ofwboot in it.
not use real locales, but this at least allows single byte character
locales to work as expected (i.e. the existing translations for sysinst).
Once we get a sysinst translation that needs more, we'll have to drop this
and the castrated setlocale(), and go for a real solution.
old, now removed sendmail. If yes, display a warning menu that offers to
replace it by a current version that selects postfix.
Thank much to Martin Husemann for extensive testing and suggestions.
XXX translations need updating.
the card's current media type. Otherwise it always defaults to 'autoselect'
which is not supported by all cards (e.g. cs(4) in NetBSD/shark).
This was causing problems when configuring an interface with dhclient, as
sysinst could incorrectly reset the media type (the net_media variable) to
'autoselect', writing incorrect ifconfig.<if> files in the target file
system.
ramdisks, and provide untested minimal implementations of mbsrtowcs()
and wcsrtombs().
Verified to fix build problems for at least the amd64 and hp300 ramdisks.
for one more that is needed, and commented out partial implementations for
the test of the functions. This should stop the re-defined symbol errors
for now.
can load a kernel from its file system.
Also tell the user what command he needs to run from the OpenFirmware shell
to configure automatic booting of the system.
Closes PR port-shark/35576.
({v,}syslog_{r,ss}) and the corresponding openlog_r/setlogmask_r/closelog_r
functions.
This should allow sparc64 to build its ramdiskbin again.
Reviewed by christos@
newfs and mount the target filesystem.
In Hauke's words: "Basically, the mac68k md_make_bsd_partitions() function,
which replaces the MI make_bsd_partitions() because the mac68k port does not
have an on-disk BSD disklabel, neglected to sign up BSD partitions for
newfs'ing and mounting by setting the appropriate flags."
Revert the commit to the install notes (rev. 1.14 of notes/mac68k/prep) which
added a warning and a workaround for this bug, as requested by hubertf in the
PR log.
"yes", and abort if it is "no" - not vice versa.
XXX - when returning to the utilities menu loop, the menu message should
be output again, I think. I couldn't figure out how that is supposed to
be done though.
default the partition to FFSv2 (instead of FFSv1).
This makes update installs add the correct bootstrap code.
Fixes PR/33682 and PR/32636 (and 33228 which has alrady been closed
as a duplicate of 32636).
I think the code it correct - but since it is only relevant when setting
the default boot partition (for mbr_bootsel) to an extended partition
it probably doesn't matter too often!
1) Add an md_post_extract() function. This function is called after
extracting the sets, and allows the arch to do something at that time.
In the case of prep, it is much easier to install the bootcode after all
the sets are extracted, so we do it in md_post_extract(). Added empty
md_post_extract() functions to all other arches so they compile.
2) Add md_mbr_use_wholedisk() and md_check_mbr(). In edit_mbr() I have
split off the code that uses the whole disk for NetBSD, into the
mbr_use_wholedisk() function. On most ports that use mbr.c, I made
md_mbr_use_wholedisk() just call that and return. On prep we create the
magical prep boot partition here. The md_check_mbr() function allows the
arch to add additional checks after the user had manually edited the MBR
to make sure the choices he made allow NetBSD to function. Added a dummy
routine to all mbr.c using arches.
3) Added code to bsdlabel.c to create a partition of type boot if
PART_BOOT is defined, but BOOT_SIZE is not defined. Also added two more
globals "bootsize" and "bootstart" which must be seeded in order to do
so. This is done on prep in md_check_mbr().
4) Added MBR_PTYPE_PREP to the list of MBR partitions.
5) Made the prep port actually install sanely. It now creates a prep
boot partition, labels it correctly, installs all the sets, and then runs
mkbootimage and dd's the bootimage into the prep partition. The result
is a prep installer that creates a bootable NetBSD installation
automatically.
6) Edited the prep menus and messages files to add new labels. In the
case of the translated files, I just added the words in english for
someone to translate later.
I tried to xcompile a few arches to make sure I didn't break anything,
but I could have missed something. Please let me know if I have broken
your arch in any way. I'll watch the autobuilds for the next few days
too. For all ports other than prep there should be no functional changes
at all.
Ask whether the user wants to delete the fetched files (ftp and floppy) after
install. Delete after sucessful install of each file.
Kill bufferred tty input in a couple of places when typeahead will be wrong.
Code added to run.c to do a 'cd' in the child, unused (so far).
for the location of the source of the sets.
Put the menu in the place on the screen that used to contian a display
of what was selected.
Now these screen work the same way as the disk partitioning ones.
Leaves more screen real-estate for more questions.
asks you where to load stuff from.
Initialise stuff so that 'install default set of sets' works.
Fix install loop, removing infinite loop if/when first kernel not selected.
of the sets after an error.
A side effect is that it now does fetch-install-fetch-install..., not
fetch-...-fetch-install-...install.
The set selection code now uses a byte per set (not a bit-pattern) removing
the restriction of 32 sets, and allowing a per set status display (TBC).
Also TBC is deleting sets from /usr/INSTALL.
However I believe it works...
systems where the bios doesn't support LBA reads.
Due to faked geometries the last few (over 30000!) sectors may not be reported
as being readable, but we expect them to be in swap. So we don't force a
root+usr install in that case.
selection (eg selecting the X sets).
libcurses got changed earlier in the year so that the erase (and refresh)
of the message window caused all the unchanged spaces to be redrawn
overwriting most of the main set selection window.
non-reentrant version here in order to avoid linker conflicts when
the rest of this file is attempted linked together with libc's
getgrent.o.
Reviewed by christos.
Since all the x_<prog> get build into crunched binaries (which does a
reachover build these days) I suspect this is entirely unnecessary.
But having deleted x_dmesg we must ensure it doesn't go there.
It is worth remembering that trailing spaces are significant in the message
files - they cause a line break if the text is auto-filled to the RH margin.
(This is often the case when the text is a format string.)
Note: nothing (yet) checks the number (or type) of the format effectors in
the message strings. I found and fixed one big error, but others will
cause core dumps.
